CRI iAtlas

This is an R Markdown Notebook. When you execute code within the notebook, the results appear beneath the code.

Try executing this chunk by clicking the Run button within the chunk or by placing your cursor inside it and pressing Ctrl+Shift+Enter.

First we are going to install the iAtlas-modules package to gain access to plotting functions.

packages = c("magrittr", "wrapr", "dplyr", "feather", "tidyr")

sapply(packages, function(x) {
  if (!require(x,character.only = TRUE))
    install.packages(x)
    library(x,character.only = TRUE)
})

Loading required package: magrittr
Loading required package: wrapr
Loading required package: dplyr

Attaching package: ‘dplyr’

The following object is masked from ‘package:wrapr’:

    coalesce

The following objects are masked from ‘package:stats’:

    filter, lag

The following objects are masked from ‘package:base’:

    intersect, setdiff, setequal, union

Loading required package: feather
Loading required package: tidyr

Attaching package: ‘tidyr’

The following objects are masked from ‘package:wrapr’:

    pack, unpack

The following object is masked from ‘package:magrittr’:

    extract

$magrittr
[1] "magrittr"  "stats"     "graphics"  "grDevices" "utils"     "datasets"  "methods"   "base"     

$wrapr
[1] "wrapr"     "magrittr"  "stats"     "graphics"  "grDevices" "utils"     "datasets"  "methods"   "base"     

$dplyr
 [1] "dplyr"     "wrapr"     "magrittr"  "stats"     "graphics"  "grDevices" "utils"     "datasets"  "methods"   "base"     

$feather
 [1] "feather"   "dplyr"     "wrapr"     "magrittr"  "stats"     "graphics"  "grDevices" "utils"     "datasets"  "methods"   "base"     

$tidyr
 [1] "tidyr"     "feather"   "dplyr"     "wrapr"     "magrittr"  "stats"     "graphics"  "grDevices" "utils"     "datasets"  "methods"   "base"     

# and our iatlas package from github
if (!require(iatlas.modules)) {
  devtools::install_github("CRI-iAtlas/iatlas.modules")
  library(iatlas.modules)
}

Loading required package: iatlas.modules

We have a collection of helper functions in the ‘notebook_functions.R’ file.

The main plotting function we’re using here is called ‘plotly_bar’. You can get help with ‘?plotly_bar’.

Here we’ll use source to bring them in.

source('functions/notebook_functions.R')

Tissue compartment fractions barplot

In this case we’ll assume you have leukocyte, stromal, and immune tissue fractions for each sample.

We’ll need a data.frame with the following columns:

Group ## (could be anything, but immune subytpe is a good choice.) fraction_type ## c(leukocyte_fraction, Tumor_fraction, Stromal_fraction) fraction ## a numeric value.

Here’s a example simulation.

df <- data.frame(
  GROUP = sample(x=c('C1','C2','C3','C4','C5','C6'), size=100, replace=T),
  fraction_type = sample(c('leukocyte_fraction', 'Tumor_fraction', 'Stromal_fraction'), size=100, replace=T),
  fraction=rnorm(n=100, mean = 0.5, sd = 0.05)
)

head(df)

The notebook_barplot function takes a very specific format, so we’ll transform the data.frame from above using the ‘build_cellcontent_barplot_df2’ function. It essentially gives it a tidy format with generic column names.

Now we’re ready to plot. We will use the plot functions from the iatlas.modules package.

Here we’re plotting the mean fraction per group with error bars.

notebook_barplot(
  df,
  sort_by_var_choice = "Group", 
  reorder_func_choice = "None",
  show_error_bars = TRUE,
  ylab = 'Fraction Mean', 
  xlab = 'Fraction Type by Group',
  title = 'Tissue type fractions by group'
)

Registered S3 method overwritten by 'data.table':
  method           from
  print.data.table     
Registered S3 methods overwritten by 'htmltools':
  method               from         
  print.html           tools:rstudio
  print.shiny.tag      tools:rstudio
  print.shiny.tag.list tools:rstudio
Registered S3 method overwritten by 'htmlwidgets':
  method           from         
  print.htmlwidget tools:rstudio

Cell content plots

Now, suppose we have cell type contents for each sample with groups, meaning each sample is assigned a group, such as immune subtype.

In the simulation below, we have 6 goups, and a number of aggregated cell types within each group. Then we have a fraction for each cell type, where the sum is ~1.0.

get_fractions <- function(n){
  f <- runif(n=6)
  f/sum(f)
}

df <- data.frame(
  GROUP = unlist(sapply(1:6, function(i) rep_len(paste0('C', as.character(i)), 6), simplify = F)),
  fraction_type = rep_len(c('Dendritic_cells', 'Eosinophils', 'Lymphocytes', 'Macrophage', 'Mast_cells', 'Neutrophils'), 6),
  fraction=unlist(sapply(1:6, function(i) get_fractions(6), simplify = F))
)

head(df)

The barplot function takes a very specific format, so we’ll transform the data.frame from above using the ‘build_cellcontent_barplot_df2’ function. It essentially gives it a tidy format with generic column names.

notebook_barplot(
  df,
  sort_by_var_choice = "Group", 
  reorder_func_choice = "None",
  show_error_bars = NA,
  ylab = 'Cell Type Fraction', 
  xlab = 'Fraction by Group',
  title = 'Celltype fractions by group'
)

Error in if (show_error_bars == TRUE) { : 
  missing value where TRUE/FALSE needed

Plot your results with iAtlas results.

# read in the "feature matrix" (fmx)
iatlas_fmx <- feather::read_feather('data/fmx_df.feather')
iatlas_fmx[1:5, 1:5]

NA

Let us then create a new data.frame that can be merged with the simulated data above. Since we’re grouping the barplots, we want to make sure the iatlas groups have different names compared to your groups… even if they represent fundementaly the same thing (i.e. C2).

# first we'll subset to a single cancer type.
kich_fmx <- iatlas_fmx %>% dplyr::filter(Study == 'KICH')

# then we'll annotate the iatlas labels.
kich_fmx$Subtype_Immune_Model_Based <- sapply(kich_fmx$Subtype_Immune_Model_Based, function(x) paste0('iAtlas_',x))

# Next we work towards formating it in such a way as to bind to the above data.frame
# Above we have cell types:
# 'Dendritic Cells', 'Eosinophils', 'Lymphocytes', 'Macrophages', 'Mast Cells', 'Neutrophils'
kich_fmx2 <- kich_fmx %>% 
  dplyr::select(Subtype_Immune_Model_Based, 
                 Dendritic_cells.Aggregate1,
                 Eosinophils.Aggregate1,
                 Lymphocytes.Aggregate1,
                 Macrophage.Aggregate1,
                 Mast_cells.Aggregate1,
                 Neutrophils.Aggregate1
                 )

# rename the cell types by removing the 'aggregate1'
colnames(kich_fmx2) <- unlist(sapply(colnames(kich_fmx2), function(x) strsplit(x, '.Aggregate1', fixed=T)))
colnames(kich_fmx2)[1] <- "GROUP"

# and finally we'll go from wide to a narrow matrix
kich_fmx2_long <- tidyr::gather(kich_fmx2, fraction_type, fraction, Dendritic_cells:Neutrophils, factor_key=TRUE)

# merge dfs
new_df <- rbind(df, kich_fmx2_long)

With our data.frame combining iAtlas and simulated results, we can make a plot showing both the new samples with iAtlas KICH results.

notebook_barplot(
  new_df,
  sort_by_var_choice = "Group", 
  reorder_func_choice = "None",
  show_error_bars = NA,
  ylab = 'Cell Type Fraction', 
  xlab = 'Fraction by Group',
  title = 'Celltype fractions by group'
)

NA

Here, we’ll use the reordering function, by selecting one of the cell types (Neutrophils) and a function (Min, Max, Mean).

notebook_barplot(
  new_df,
  sort_by_var_choice = "Neutrophils", 
  reorder_func_choice = "Max",
  show_error_bars = NA,
  ylab = 'Cell Type Fraction', 
  xlab = 'Fraction by Group',
  title = 'Celltype fractions by group sorted by Neutrophils'
)

NA